AMD Radeon RX 580 vs NVIDIA GeForce GTX 1660 Comparison

The NVIDIA GeForce GTX 1660 and AMD Radeon RX 580 represent two distinct approaches to mid-range graphics from different architectural eras. The data shows a decisive sweep: the GTX 1660 wins all 10 head-to-head benchmark comparisons, yet the RX 580 holds a higher average benchmark score and a better percentile ranking. This apparent contradiction is the core investigative question, and the answer lies in how each card’s raw specifications translate into real-world test results. The verdict is clear for raw performance: the GTX 1660 is the superior choice, but the RX 580’s larger memory pool and higher overall average score suggest it may have specific strengths that the head-to-head tests do not fully capture.

The Verdict

Based strictly on the benchmark data, the NVIDIA GeForce GTX 1660 is the definitive winner for users prioritizing raw compute and graphics performance in the tested workloads. It outperforms the AMD Radeon RX 580 in every single head-to-head comparison, with particularly large margins in DirectX 9 (42.7% faster), GPU compute (42.3% faster), and DirectX 10 (32.6% faster). The GTX 1660 also shows a significant lead in the modern 3DMark Steel Nomad DX12 test, scoring 1065 versus 1005, a 6% advantage. For gamers and professionals using DirectX or Vulkan APIs, the GTX 1660 is the clear pick from this data.

However, the AMD Radeon RX 580 is not without its own merits, as evidenced by its higher average benchmark score of 12928 compared to the GTX 1660’s 11680. This places the RX 580 in the 53rd percentile of all GPUs, while the GTX 1660 sits at the 51st percentile. The RX 580’s nearest rivals include the NVIDIA GeForce GTX 1660 SUPER, which it trails by a mere 0.4%, suggesting it remains competitive within its performance class. The RX 580 should be the choice for users who value a larger memory buffer (8 GB versus 6 GB) and higher raw FP32 throughput (6.175 TFLOPS versus 5.027 TFLOPS), even if the benchmark results do not reflect this advantage in the tested scenarios.

The GTX 1660’s wins are comprehensive, but the RX 580’s higher average score indicates that in a broader set of applications, it may perform differently. The data suggests the GTX 1660 is better for users running the specific benchmarks listed, while the RX 580 may excel in other, untested workloads that favor its memory bandwidth and compute capacity. Ultimately, the GTX 1660 wins the head-to-head, but the RX 580 remains a viable alternative for specific use cases.

Architecture Differences

The two cards are built on fundamentally different architectures, which explains their divergent benchmark results. The NVIDIA GeForce GTX 1660 uses the TU116 chip based on the Turing architecture, manufactured on a 12 nm process at TSMC. This chip packs 6,600 million transistors onto a 284 mm² die, yielding a transistor density of 23.2 million per mm². In contrast, the AMD Radeon RX 580 employs the Polaris 20 chip based on the older GCN 4.0 architecture, built on a 14 nm process at GlobalFoundries. The Polaris 20 contains 5,700 million transistors on a 232 mm² die, with a slightly higher transistor density of 24.6 million per mm².

The GTX 1660’s Turing architecture features 1408 shading units, 88 texture mapping units, and 48 render output units. The RX 580 counters with significantly more shading units (2304) and texture units (144), but fewer ROPs (32). This difference in core configuration explains the RX 580’s higher theoretical texture rate of 193.0 GTexel/s versus the GTX 1660’s 157.1 GTexel/s. However, the GTX 1660 achieves a much higher pixel rate of 85.68 GPixel/s compared to the RX 580’s 42.88 GPixel/s, despite having more ROPs on paper. This suggests the Turing architecture is more efficient at pixel processing per ROP.

Memory architecture is another key divergence. The GTX 1660 uses 6 GB of GDDR5 on a 192-bit bus, providing 192.1 GB/s of bandwidth. The RX 580 offers 8 GB of GDDR5 on a wider 256-bit bus, delivering 256.0 GB/s. The RX 580’s superior memory bandwidth is a direct result of its wider bus, which could benefit high-resolution textures and compute workloads. Additionally, the RX 580 has a higher FP32 throughput of 6.175 TFLOPS versus the GTX 1660’s 5.027 TFLOPS, and it handles FP16 at a 1:1 ratio (6.175 TFLOPS), while the GTX 1660 uses a 2:1 ratio (10.05 TFLOPS). The GTX 1660 also supports a higher DirectX feature level (12_1 versus 12_0) and a newer Vulkan version (1.4 versus 1.3).

FAQ

Q: Which card has a higher average benchmark score?

A: The AMD Radeon RX 580 has a higher average benchmark score of 12928, compared to the NVIDIA GeForce GTX 1660’s 11680. This places the RX 580 in the 53rd percentile of all GPUs, while the GTX 1660 sits at the 51st percentile.

Q: How large is the GTX 1660’s lead in the head-to-head benchmarks?

A: The GTX 1660 wins all 10 head-to-head comparisons. The biggest margins are in Passmark DirectX 9 (42.7% faster), Passmark GPU Compute (42.3%), and Passmark DirectX 10 (32.6%). The smallest lead is in Passmark G2D, where it is only 0.9% ahead.

Q: Does the RX 580 have any advantages in memory configuration?

A: Yes. The RX 580 has 8 GB of memory on a 256-bit bus, providing 256.0 GB/s of bandwidth. The GTX 1660 has 6 GB on a 192-bit bus, with 192.1 GB/s. The RX 580’s larger memory and higher bandwidth are notable differences.

Q: Which card has a higher theoretical compute throughput?

A: The RX 580 has a higher FP32 throughput of 6.175 TFLOPS, while the GTX 1660 delivers 5.027 TFLOPS. The GTX 1660, however, achieves 10.05 TFLOPS in FP16 using a 2:1 ratio, whereas the RX 580 is capped at 6.175 TFLOPS for FP16.

Q: How do the two cards compare in terms of power requirements?

A: The GTX 1660 has a TDP of 120 W and a suggested PSU of 300 W, while the RX 580 has a TDP of 185 W and a suggested PSU of 450 W. Both use a single 8-pin power connector.

Q: What is the release date difference between the two cards?

A: The NVIDIA GeForce GTX 1660 was released on 2019-03-13, while the AMD Radeon RX 580 was released earlier on 2017-04-17. Both are now end-of-life products.

Specification Differences

The two cards differ across nearly every major specification category. The GTX 1660 uses a 12 nm process, while the RX 580 uses a 14 nm process. The GTX 1660’s chip, TU116, has 6,600 million transistors on a 284 mm² die, versus the RX 580’s Polaris 20 with 5,700 million transistors on a 232 mm² die. The GTX 1660 has a higher base clock of 1530 MHz and boost clock of 1785 MHz, compared to the RX 580’s 1257 MHz base and 1340 MHz boost.

Memory is a major difference: the GTX 1660 offers 6 GB of GDDR5 on a 192-bit bus with 192.1 GB/s bandwidth, while the RX 580 provides 8 GB on a 256-bit bus with 256.0 GB/s. The GTX 1660 has 1408 shading units, 88 TMUs, and 48 ROPs, versus the RX 580’s 2304 shading units, 144 TMUs, and 32 ROPs. The GTX 1660 achieves higher pixel rate (85.68 GPixel/s) but lower texture rate (157.1 GTexel/s) compared to the RX 580 (42.88 GPixel/s and 193.0 GTexel/s). FP32 performance favors the RX 580 (6.175 TFLOPS) over the GTX 1660 (5.027 TFLOPS). The GTX 1660 has a lower TDP of 120 W versus 185 W, and a lower suggested PSU of 300 W versus 450 W. Display outputs also differ: the GTX 1660 has 1x DVI, 1x HDMI 2.0, and 1x DisplayPort 1.4a, while the RX 580 has 1x HDMI 2.0b and 3x DisplayPort 1.4a. The GTX 1660 supports DirectX 12_1, while the RX 580 supports DirectX 12_0. The launch MSRP for the GTX 1660 was 219 USD, and for the RX 580 it was 229 USD.

Head-to-Head Benchmarks

The head-to-head results are a complete sweep, with the GTX 1660 winning all 10 tests. The most dramatic victory is in Passmark DirectX 9, where the GTX 1660 scores 177 against the RX 580’s 124, a 42.7% advantage. This is followed closely by Passmark GPU Compute, where the GTX 1660’s 4963 score beats the RX 580’s 3488 by 42.3%. The Passmark G3D test shows a 32.1% lead for the GTX 1660, with scores of 11646 versus 8813. The GTX 1660 also dominates in Passmark DirectX 10 (61 vs 46, a 32.6% lead) and DirectX 11 (79 vs 60, a 31.7% lead). In Geekbench OpenCL, the GTX 1660 scores 47850 against 37453, a 27.8% advantage. The Geekbench Vulkan test shows an 11% lead for the GTX 1660 (50137 vs 45173). The 3DMark Steel Nomad DX12 test has the GTX 1660 ahead by 6% (1065 vs 1005). The smallest margin is in Passmark G2D, where the GTX 1660 leads by just 0.9% (776 vs 769). In DirectX 12, the GTX 1660 wins by 14% (49 vs 43).

These results reveal a pattern: the GTX 1660’s advantage grows in older API tests (DirectX 9 and 10) and compute workloads, while its lead narrows in more modern tests like 3DMark Steel Nomad and G2D. This suggests the Turing architecture is particularly efficient at legacy compatibility and compute tasks, while the RX 580’s GCN architecture may be more competitive in certain modern scenarios, even though it still loses.

Where Each One Wins

The GTX 1660 wins in every benchmark category tested, but the magnitude of its victories varies, indicating where each card is strongest. The GTX 1660 is the clear winner for users running legacy DirectX 9 or DirectX 10 applications, as it shows leads of 42.7% and 32.6% respectively. It is also the better choice for GPU compute tasks, with a 42.3% lead in Passmark GPU Compute, making it more suitable for general-purpose compute workloads like OpenCL-based applications. For Vulkan gaming, the GTX 1660 is 11% faster, and for DirectX 12, it maintains a 14% edge. The GTX 1660 also dominates in the overall Passmark G3D score, being 32.1% faster, which indicates it is the superior choice for most 3D gaming scenarios.

The RX 580, despite losing all head-to-head tests, has specific strengths that are not reflected in the win column. Its higher average benchmark score of 12928 versus the GTX 1660’s 11680 suggests that in a broader benchmark suite, it could outperform the GTX 1660. The RX 580’s nearest rival, the GTX 1660 SUPER, is only 0.4% faster, indicating the RX 580 is competitive with a card that is a direct successor to the GTX 1660. The RX 580’s 8 GB memory and 256 GB/s bandwidth make it theoretically better for high-resolution textures and memory-intensive workloads. Its higher FP32 throughput of 6.175 TFLOPS could benefit compute applications that rely on single-precision floating-point, even though the Passmark GPU Compute test shows the GTX 1660 winning by a large margin. The RX 580’s higher percentile ranking (53rd versus 51st) also suggests it may perform better in a wider range of tests than the head-to-head selection shows. For users who prioritize memory capacity and theoretical compute power over the tested benchmarks, the RX 580 remains a relevant option.